Comprehensive Evaluation and Comparison of Machine Learning Methods in QSAR Modeling of Antioxidant Tripeptides

Abstract

Listen

Translate article

Due to their multiple beneficial effects, antioxidant peptides have attracted increasing interest. Currently, the identification of antioxidative peptides and screening bioactive peptides are based on wet-chemistry methods which are time-consuming and highly rely on many advanced instruments and trained personnel. Compared to wet chemistry methods for preparation and screening bioactive peptides, quantitative structure-activity relationship (QSAR) analysis as an in silicon method can be more efficient and cost-effective. However, model performance of QSAR studies on antioxidant peptides was still poor due to the difficulties from data set division to regression methods. The objective of this study was to compare most popular and promising machine learning methods for antioxidant activity modeling and screening of tripeptides and identify the critical amino acid features that determine the antioxidant activity. We adopted 553 numerical indices of amino acids to characterize 130 tripeptides with known antioxidant activity from published literatures, and then six advanced feature selection methods plus pairwise correlation were used to screen the most important indices for antioxidant activity and model building. Fourteen machine learning methods were used to build models based on the six feature selection methods, respectively. Among the 84 models, the best model with R2Test of 0.847 and MSETest of 0.393 for tripeptide antioxidants was obtained based on FI-RFR plus XGB. Based on the predicted antioxidant values of 7870 unknown tripeptides, the potential high antioxidant activity tripeptides all have a tyrosine, tryptophan, or cysteine at the C-terminal position. In addition, results also showed that C-terminal amino acids contributed the most to antioxidant activity, while the central amino acid contributed the least. Non-linear regression methods were more suitable for QSAR study on antioxidant activity.